Every firm has a different marketing objective: some look for maximum ROI and some are willing to spend to break even in hopes that future orders will drive the profit, while others see search as a means of spreading the brand message.
Whatever your objective, you’re more likely to reach it the more tightly your success metric jibes with the real value of the traffic driven.
Most retailers sell products in an array of different categories with varying AOVs and margin structures. If a retailer says they’re comfortable spending 25% of sales revenue on marketing, does that go for the electronic widgets where 80% of the revenue pays for the cost of goods? What about the cable widgets for which only 30% of the revenue goes to the wholesaler?
Consider the following chart. In this simplified world the retailer sells high ticket (HT) items for $300 and low ticket (LT) items for $50. Some products have high margins (HM) at 65%; others have low margins (LM) at 30%. This creates four subclasses of sales: high-ticket/high margin (HTHM), high-ticket/low margin (HTLM), low-ticket/high margin (LTHM), and the ever popular low-ticket/low margin (LTLM). The retailer bids by margin aiming to break even against the advertising costs.
This retailer is rightly pleased with the way the program is going at this level of granularity.
What’s instructive here is that if the retailer were looking instead at Cost/Sales as a proxy, and bidding to achieve that 41% ratio which makes the numbers work on average, their system would bid up the HTLM and LTLM items and bid down the LTHM and HTHM items.
Conversely, bidding to achieve a CPO objective of $35 would mean bidding down the HTLM and HTHM related keywords and bidding up keywords related to the LTLM products.
As a proxy, some companies aim at different cost to sales targets (or revenue multipliers) based on the keyword’s related product category. That’s not a bad approach, but given that people don’t always buy what they searched for, and often buy from more than one category in a given shopping cart, it’s a poor proxy at best.
Moreover, as any retail veteran knows, some products have higher fraud, cancel and return rates, which can make the demand sales look better than they are. Sale events can also make PPC advertising look artificially efficient, when in fact margins are thinner.
There are three good ways to approach margin bidding:
Feeding on the fly. If your website has the margin associated with each product available at checkout, that data can be passed to your bidding system in place of or in addition to the sales dollars. Advantages: Happens dynamically and in a timely fashion, no need for additional processes, clean and simple. Disadvantages: Doesn’t help with high fraud/cancel rate products, and some websites don’t have the information available at checkout.
Backfeeding. A nightly process to match up orders to the margins associated with them. Advantages: Gives retailers the ability to drop frauds and cancels. Disadvantages: Delay between the sales events and the data, which can be problematic for highly seasonal retailers. Matching up the data and sucking in ftp files can be messy, flaky and brittle, chewing up valuable IT resources.
Hybrid. The best of the best, feeding the margin data on the fly coupled with a nightly process to knock out frauds and cancels.
Once the discussion is about margin rather than proxies for value, some interesting and powerful techniques can be applied with much clearer vision. For the company driving strictly for ROI, using PPC advertising as a cash machine, consider the power of margin discussions: you’re no longer driving by percentages, but by dollars.
Aiming for a 50% cost to margin ratio may generate less net revenue (margin – ad costs) than aiming for a 75% cost to margin ratio. Would you rather keep 50% of $1,000, or 25% of $5,000? The additional aggressiveness of higher efficiency thresholds may actually improve profits, improving the top line and bottom line simultaneously.
Whichever system your team can execute, tying performance to the right numbers makes cents…and dollars!
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.